One of the largest and most recognized health systems in New England is tackling some of its biggest challenges with an innovative approach to AI. With 11 hospitals, nearly 120 primary care and specialty practices, and over 6,000 providers, Beth Israel Lahey Health (BILH) is no stranger to the hurdles healthcare is facing today: burnout, recruitment challenges, reimbursement pressures, degrading patient experience, and more. Improving the workforce experience is a strategic priority for BILH, leveraging innovative technologies including ambient AI scribes.
BILH chose a physician- and patient-centric approach that is contrary to the usual strategy of opting first for a solution that easily integrates with the EHR. This allowed physicians to drive adoption in care settings, specialties, and encounter types that worked best for them without being constrained by product capabilities. While scoping the project and looking at options, several physicians approached system leadership to advocate for a solution they were already using—and did not want to give up. That solution was Heidi's AI scribe.
After a 6‑month trial where Heidi was offered to 1,000 providers, a majority had adopted the technology.
In a survey of the users
89% said that they were satisfied with the quality of the note
88% said that Heidi captured complex medical terms
90% said they felt more present and made more eye contact with their patients when using the tool
82% said that they experienced reduced cognitive load during their visit
Addressing burnout created by burdensome documentation and administrative tasks is a priority for most health systems. These results show that ambient AI can deliver relief that benefits providers, patients, and the system at large.
Nearly half of all physicians in the US report burnout. Many also report that they are considering leaving the profession, and the current turnover rate among clinicians is over 7%. Fifty‑seven percent of providers report that the burden of documentation is a major cause of their burnout. This burden on physicians translates to patients, who may wait over a month for an appointment and say their providers spend too much time in the EHR and not enough time engaging in the visit. Provider organizations feel the impact on their bottom line in the high cost of turnover and in poor patient experience and outcomes.
"I've used [other AI scribes] and I'm amazed by how much better Heidi is in terms of formulating the thought process." - Physician, Emergency Department
BILH deployed a standalone AI scribe tool, independent of their EHR, which empowered physicians to creatively explore unconventional use cases across specialties, care settings, and document types—including multilingual documentation—beyond typical AI scribe applications. Given the extraordinarily positive feedback received from the initial providers, BILH is in the process of rolling out Heidi to all their clinicians and integrating Heidi with Epic.
Introduction
Physicians are drowning in documentation and patient‑care‑related administrative tasks like referral letters. A report from the Century Foundation that aggregated a number of studies related to physician burnout reported:
"Time spent on electronic health records (EHR) is also a major stressor for physicians. Many physicians report that EHR systems can be challenging to use, sometimes leading to patient harm, and also that payers require significant documentation. Estimates of physician time spent on the EHR varies, but one study found that primary care physicians spend about sixteen minutes per patient visit on an EHR. Given the focus on volume in the health care system, it is easy to imagine why this time on EHRs would be a major source of stress for physicians. In one physician survey, 75% of respondents attributed burnout to EHRs. Again, this focus on time on the electronic health records further detracts from time with patients and colleagues."
AI scribes have the potential to significantly reduce not only documentation and note‑taking time, but also time spent on many other tasks like referral letters.
While the research on this new technology is still in its infancy, there is enough data to say with confidence that AI scribes save physicians time. By transcribing the visit, providers can stay focused on the patient while avoiding hours of documentation time outside of patient visits. One study showed that on average providers spent three hours less time on after‑hours documentation a week.
For an AI scribe to deliver transformative value, it needs to be adopted by at least 60% of the clinicians across the organization.
High adoption is what Beth Israel Lahey Health is trying to achieve in its partnership with Heidi. BILH achieved this with an implementation strategy that focused on communication, training content, and peer videos. The results from the first six months indicate they are steadily working towards achieving this goal.
"I've never been happier in my job as a staff physician… Simple 'Dragon' dictation has always been helpful but is not very dependable (i.e. prone to errors), resulting in excessive proofreading time. Human scribes are only slightly more helpful than dictation, and they typically are not 'thinking like doctors' when they are typing the note…which ultimately ends up too basic to capture the visit in necessary detail. Heidi is a revolutionary tool that overcomes all these limitations." - Physician, Rheumatology
This is no small feat given that BILH consists of 155 facilities, including 14 community and specialty hospitals, 11 urgent care centers, 90 primary care practices, and 28 specialty practices. There are more than 4,700 physicians and 39,000 employees across these diverse healthcare settings.
Method
Three cohorts were onboarded between January and June of 2025. By the end of the trial period, 1,000 clinicians had been offered access to Heidi and 760 signed up to use Heidi's AI scribe.
The alpha phase was launched to 100 clinicians in mid-January 2025. The 4-week alpha phase included 50 clinicians from Lahey Hospital and 50 clinicians from BILH's other locations, across 25 specialties.
The beta phase added 500 clinicians who tested the software for another 8 weeks.
Finally, the gamma test was rolled out to include other physicians who were referred or showed interest in using the AI scribe. This group included 47 specialties.
Participants were offered training on using Heidi's AI scribe, before commencing the trial period. As part of the training process, BILH and Heidi offered clinicians:
Department‑specific training materials, including customized workflow videos created in collaboration with BILH providers
Provided custom templates (500+ templates created across nearly 50 specialties).
12 office hours a week.
Usage data was collected in partnership with Heidi Health to understand the engagement with the tool during the study period.
The use of the AI scribe was not mandatory and entirely at the providers’ discretion. Each session with the AI scribe corresponded to one visit with a single patient. Participants were asked to review and approve all documentation produced by the AI scribe to ensure accuracy and quality of documentation.
"Heidi has made clinic less stressed, draining, and frantic. I feel liberated from the computer and connected to patients." - Physician, Urology
Providers were briefed that the quality and accuracy of all documentation produced within the trial remained their responsibility.
The study was designed to assess three primary outcomes regarding the impact of an AI scribe on BILH’s providers.
Primary Outcomes
Impact of the AI scribe on physicians’ administrative burden—namely, reducing the time taken to write clinical notes and prepare other clinical documentation and the overall time taken to complete clinics.
Impact of the AI scribe on cognitive load and physician patient engagement.
Secondary Outcomes
Assess the acceptability of AI scribe technology with patients.
Impact of the AI scribe on physicians’ documentation quality.
"I love the way using Heidi has allowed me to talk to the patient face to face, listen and engage with them, think things through because I'm truly focused on what they're saying and not typing and trying to get it down. My visits are so much more satisfying for me, and I'm very curious if patients have noticed that too." - Provider, Primary Care
Participants were surveyed at the end of the trial period to provide feedback on the AI scribe, using Likert scale to agree or disagree with a series of statements that aligned to the trial’s primary and secondary outcomes. Additionally, key metrics like minutes of recordings transcribed, number of consultations transcribed, frequency of use, and significant adverse events were tracked throughout the trial.
Results
Based on the results of the initial testing phases, BILH is now pursuing launching a full integration between the Heidi AI scribe and Epic to further streamline their provider workflows using Heidi.
Between mid-January and early June 2025, a survey of BILH providers using Heidi’s AI scribe showed:
1171 users
47 specialties
Nearly 110,000 encounters
Generating over 18,000 notes and other documents per week
Over 1,500 encounters in other languages
89% said that they were satisfied with the quality of the note
88% said that Heidi captured complex medical terms
90% said they felt more present and made more eye contact with their patients when using the tool
74% said the time they spent outside work hours (pajama time) had been reduced
82% said that they experienced reduced cognitive load during their visit
"I used Heidi with an aggressive dementia patient that I was seeing in the nursing home, I then spoke to the patient's daughter who changed code status and then spoke to social work. Heidi did a phenomenal job capturing a complicated patient visit and family meeting. [It] has been very helpful in the nursing home and sub‑acute rehab setting." - Provider, Extended Care Community Program
The clearest indicator of success is that BILH is still steadily adding users every week, primarily through word of mouth. They continue to receive unsolicited feedback from providers daily about the impact the tool has made on their lives as well as the creative ways they are using it.
"I have been so impressed with Heidi. I have been telling many of my colleagues how much of a game changer this is for documentation. One of the best aspects of Heidi is our ability to devote all our attention to the patient while they are talking. I no longer take copious notes while they are telling me their medical history. The HPI generated from the transcript is succinct and accurate—even better than if I had dictated the note.” - Provider, Gastroenterology
Conclusion
AI will play a key role in addressing the issues facing healthcare today. To achieve true transformation in the future requires high adoption of the solutions implemented today. Heidi has developed a solution designed to deliver the highest provider adoption. The results at BILH showcase the success of Heidi's approach, which:
Understands that note writing is a complex workflow that requires a multi-faceted approach to streamlining documentation.
Approaches adoption as the key to success and provides the tools, features, and flexibility that providers want.
Sees integration as a feature that can be added and touches the health record in the right ways but doesn’t drive the design.
When an ambient AI solution is focused on the physician, it drives industry‑leading activation, which:
Reduces burnout and turnover by saving 5 minutes per visit and up to two hours a day in documentation time.
Brings joy back to the job for providers by reducing time spend on administrative tasks.
Improves patient experience by allowing providers to focus on the patient instead of notetaking.
Creates a framework for the future of work with AI by collecting clinical data for future use and building workflows that are more efficient.
Opens capacity by freeing enough time to add 2-4 patients a day per provider, improving patient access and outcomes.
When considering ambient AI, organizations should look for a partner that understands the impact of these solutions across the organization and a solution that is designed to achieve the adoption needed to see results like those of BILH.
